Introduction
149
IPLink Software Configuration Guide
14 • Link scheduler configuration
Introduction
This chapter describes how to use and configure the IPLink software Quality of Service (QoS) features. Refer
to chapter 19,
“Access control list configuration”
on page 211 more information on the use of access control
lists.
This chapter includes the following sections:
•
Quick references (see
page 152
)
•
Packet Classification (see
page 154
)
•
Assigning bandwidth to traffic classes (see
page 152
)
•
Link scheduler configuration task list (see
page 153
)
QoS in networking refers to the capability of the network to provide a better service to selected network traffic.
In the context of VoIP, the primary issue is to control the coexistence of voice and data packets such that voice
packets are delayed as little as possible. This chapter shows you how to configure IPLink software to best use
the access link.
In many applications you can gain a lot by applying the minimal configuration found in the quick reference
section, but read sections
“Applying scheduling at the bottleneck”
and
“Using traffic classes”
first to understand
the paradox of why we apply a rate-limit to reduce delay and what a “traffic-class” means.
Applying scheduling at the bottleneck
When an IPLink acts as an access router and voice gateway, sending voice and data packets to the Internet, the
access link is the point where intelligent use of scarce resources really makes a difference. Frequently, the access
link modem is outside of the IPLink and the queueing would happen in the modem, which does distinguish
between voice and data packets. To improve QoS, you can configure the IPLink to send no more data to the
Internet than the modem can carry. This keeps the modem’s queue empty and gives the IPLink software con-
trol over which packet is sent over the access link at what time.
Using traffic classes
The link scheduler needs to distinguish between different types of packets. We refer to those types as “traffic-
classes”. You can think of the traffic-class as if every packet in the IPLink has a tag attached to it on which the
classification can be noted. The access control list “stage” (ACL) can be used to apply such a traffic-class name
to some type of packet based on its IP-header filtering capabilities. The traffic-class tags exist only inside the
IPLink, but layer 2 priority bits (802.1pq class-of-service) and IP header type-of-service bits (TOS field) can be
used to mark a specific packet type for the other network nodes. By default the traffic-class tag is empty. Only
two types of packets are automatically marked by the IPLink software: voice packets and data packets origina-
tion from or destined to the IPLink itself are marked as “local-voice” and “local-default” respectively. Please
refer to
figure 23
on page 150 when using the ACL to classify traffic. It illustrates the sequence of processing
stages every routed packet passes. Only stages that have been installed in the data path with a “use profile...”
statement in the corresponding interface configuration are present. Both an input direction ACL on the receiv-
ing interface as well as an output ACL on the transmitting interface can be used to classify a packet for special
handling by the output link scheduler on the transmit interface. But as visible from the figure no ACL can be
used for an input link scheduler.